SOURCE DEPENDENCY OF SPECTRAL TRANSFER CHARACTERISTICS FOR REALTIME SPECTRAL FORECASTING

Authors

• Ahmed A. Torky PhD, The British University in EgyptFollow
• Susumu OHNO Ph.D., International Research Institute of Disaster Science, Tohoku University, Japan

Document Type

Book

Publication Date

2024

Abstract

By using strong motion data along the Pacific coast of Tohoku district, Japan, we investigate 1) the source location dependency of nonstationary spectral transfer characteristics between front-sites and distant target sites, and 2) source-area specific machine learning models to predict target-site spectra from front-site spectra and their prediction accuracy. The major results are 1) Front-P to target-S spectral ratios show more sensitive to source location than front-S to target-S spectral ratios, and 2) Source-area specific neural networks show better performance to predict not only front-S to target-S but also front-P to target-S cases, compared to unrestricted source area case. These results could be useful to improve the performance of EEW in accuracy and alert time.

Recommended Citation

Torky, Ahmed A. PhD and OHNO, Susumu Ph.D., "SOURCE DEPENDENCY OF SPECTRAL TRANSFER CHARACTERISTICS FOR REALTIME SPECTRAL FORECASTING" (2024). Civil Engineering. 244.
https://buescholar.bue.edu.eg/civil_eng/244